Mode decomposition and sparse reconstruction of the pressure field of the cylinder entering water

Abstract

Water entry is a difficult multiphase flow issue, characterized by intricate interconnections between gas, liquid and solid. In this study, we focus on the pressure field of the water-entry cylinder as the research object, which we consider to be a complex dynamic system. To analyse this system, we apply the data-driven method to decompose and reconstruct the system. Firstly, based on the CFD (Computational Fluid Dynamics) approach, the pressure snapshots are collected, and the dataset of the system is constructed. Then, POD (Proper orthogonal decomposition) method is employed to decompose and reconstruct the system. By analyzing the corresponding decomposition mode, we can deduce the motion state of the cylinder and its attached cavitation evolution. Finally, the sparse reconstruction algorithm is proposed, which allows us to reconstruct a simplified pressure field using a limited number of pressure measuring points (≥4). We offer several suggestions to enhance the accuracy of the algorithm: when POD mode number is close to the number of measuring points, the reconstruction accuracy is relatively high; likewise, the accuracy increases with the number of measuring points. For the assessment of the sparse reconstruction, the errors of this algorithm are largely concentrated in the area where the gas-liquid phase transitions occur. The trained placements of the measuring points are randomly determined. In addition, the algorithm exhibits good generalization and is applicable to predicting the unknown pressure field. This paper presents a fresh and innovative quantitative perspective on water entry and offers valuable insights for the design of experimental sensors.